If it's a typical sidebar widget ignore the 'talk to an ad network' suggestions. If standard IAB banner sizes aren't right for your widget (and I really don't see how you'd cram one into a widget), then ad networks have nothing for you.

Don't price this by the impression. If your advertisers are thinking CPM they'll rapidly start evaluating what this untargeted, run-of-network inventory is worth, which is likely close to zero. You won't get a $1.00 CPM if you price it by the impression. You'll be lucky to get a $0.20 CPM.

Instead, I'd try and sell a few sponsorships, Deck Network style, each sponsorship granting a share of the traffic for at least a month and for at least a four-figure sum each. The audience is looking at a Twitter widget, so your natural audience is Twitter-related businesses. Develop an unobtrusive text-based ad unit, something like the promos for Twitter clients that Twitter itself runs on twitter.com, and then pitch it to every Twitter-related business you can find.

Remember, high prices for a big share of a really relevant audience and do whatever you can to keep them from thinking about the inventory in terms of CPM, because five million ad impressions is not a hell of a lot when rationally priced. You need to do a lot of hand waving and sprinkle a lot of fairy dust ('reaching key influencers', etc.) to sell such a small amount of inventory for a meaningful amount.

One kind of ad you could run that wouldn't have trouble with cramming into a widget is popups.

Don't downvote me because you hate popups, so do I. It's a legitimate possibility that you might consider looking into. If it's a question of shutting down the service because of bleeding costs or running popups on 5% of pageviews, its an interesting question.

This is a great idea if you want to get rid of all your users. Nobody likes popups, and the people whose sites contain the widget would consider it a dick move and uninstall it immediately, I suspect.

I agree, but this also solves his paying for bandwidth problem.

Sponsorships are a fair bit of work to setup (from a sales perspective) but result in higher revenue.

Setup the widget so it gets deeply skinned with the sponsor's branding. Check out pandora.com for a nice implementation - the whole site takes on the look and color of the advertiser.

The first few sponsors are the hardest to get, but once you get going others will find their way to you and you'll be able to have steady income.

If standard IAB banner sizes aren't right for your widget then use text ads.

CPM is always the last choice. Regardless of CPM, CPC or heck even CPA, test your traffic using private run of network ads. You'll have complete control over them so you can see the honest to goodness truth and be able to relay that to the ad network sales folks for initial measurements and pre-sales.

This might be slightly deceptive... but you could try adding affiliate codes to all links to product websites (Amazon, NewEgg, etc.)

edit: URL shorteners might make this difficult, but still possible if you can keep the widget responsive enough while following links server-side.

I like this one best - ads are probably not an option because widgets have limited real estate.

To eliminate the deceptive part, just message all yours users first before changing it. Be candid about money issues (ala reddit). Start with just Amazon maybe.

"Hey guys, this thing has blown up and we can't afford the hosting cost. To keep it free, we've got to try something. So starting in 1 month, we'll automagically insert our Amazon affiliate code into any tweets than contain an Amazon link. All other links won't be touched. We really didn't want to do this but it is either this or shut down the widget (or start charging for it). Thanks for bearing with us as we run this!"

You might lose 10% of freeloaders but this is a good thing.

I've long wondered if this kind of approach is actually something that affiliate programs would allow if you got big. You are not promoting their wares, you are simply profiting off of organic promotion that would have occurred anyway. I wouldn't be surprised if Amazon or whomever decided to close your account if you started making significant amounts of money.

Deal aggregating sites do this I am pretty sure, slickdeals.com for example.

If someone clicks on a link that your widget provided (a link that would not have otherwise been there) it seems like you are actually providing value to the company.

The link would be there regardless of whether or not you (as an advertiser) ran an affiliate program.

I've actually tried this on the main site, but it was not so successful. Mostly because the links are shortened, and I don't have the capacity to unshorten every link (most computation is done client-side anyway), and programs like VigLink don't look at short links either.

Have link HTML printed out as per normal, attach JS that routes every clicked link through your servers. E.g.

    This is a twitter <a href="http://bit.ly/blah"
    onclick="document.location='http://yourservers.com/follow?url=' +
    escape(this.href);return false">link</a>.

That way you can look up the Twitter links only if they are clicked and any slow down happens after the click, not before display.

As a bonus you get to see what links everyone is clicking as well.

You'd likely need to roll your own mini-crawler to dig out the shorten links, and where applicable, create new ones and replace them. Questionable indeed.

Ahh, yeah hadn't thought about that....you'd have to unshorten and then re-encode every amazon link. Still though, seems like it might be possible.

You may want to use http://unshort.me api service to unshorten the links.

I think this might be the best way, for the widget as is. There might be a problem in that most of the search terms don't typically include a affliatable link.

I think the best way might to go to the premium model, the only thing is that by restricting the number of features might just get people to run to another service.

If not enough links are present, one could also create a map of keywords and appropriate products and create links where none exist today. (I have no idea if this would break Twitter's ToS.) Heck, advertisers may even buy certain keywords directly from you... advertisers like me. ;-)

You could be really deceptive and rewrite affilliate links on the page the widget's on. I wouldn't do it though.

Here's how I'd do it... I'd massage this a bit with some closed alpha testing, etc... but this is the gist:

Sell ads and kickback a percentage to the siteowners while working in charity donations to remind folks that you're not evil.

Create a script to categorize your install base. Eg: Gadgets, Tech, Food, Personal, etc. In version 2, plan to categorize in real-time contextually. A good ad network partnership (see below) will help you with this technology.

Change your TOS. Send an e-mail to everyone asking them to verify that your script categorized them properly, to check out the new TOS and the new ad feature. Clearly and honestly tell them why you're making this change - the project will die if you don't.

Test some "run of network" ads. Input a few different variations, measure clicks (volume and geo are important).

After a couple of weeks, approach a few different ad networks at the same time. An Adtech conference is coming up in NY in November... use their exhibitor list as a hit list.

Negotiate with the ad networks. Depending on the demographics and CTR's, I bet with a 5m install base, you'll have some good offers. Reliability is key here - ad networks start and stop all of the time. In contrast, remember that big networks come with more constraints.

Here's the part that's going to keep this from being evil: Allow your siteowners to sign up and insert their own affiliate / user codes. Give them a percentage of the ads. If someone does not sign up, give that same portion to a charity such as Kiva.

As an alternative to selling ads - start looking to sell the whole thing. There are a few funded startups I can think of that would keep your beer fridge full for a couple of years in exchange for what you have built.

As a final (tacked on edit) idea: Survey your siteowners. With such a large install base, if you ask what advanced features people would like, you may be able to get some people to upgrade to a paid subscription model. You might be able to use this in conjunction with the adrev model above. In your survey, don't give the siteowners any ideas - you don't want to sway them.

That sounds like a huge amount of work to set up.

You wanna make money but don't wanna work for it?

Setting up an ad network that charges advertisers, targets ads and then compensates site owners is not a fun way to monetize a widget. It's recreating AdSense from scratch. Think of the tax issues alone!

If you could build that ad system, coming up with 5 million impressions would be the easy part.

You would not need to build an ad network. There are a few hungry ad networks out there already that offer text ads and have systems similar to this already in place.

The only challenge really is in categorizing the sites where the widget appears (to target the ads so that the CPM ratio doesn't make it worthless). Past that, it's really just seeing who has a good base of customers buying text ads and letting them handle all of the dirty work.

Ask your users if they'd mind you including VigLink or something like it in your widget. It monetizes outbound hyperlinks that are already on the page by rewriting them to be affiliate links when there's a relevant program. It doesn't change the behavior of the site and doesn't overwrite any that are already affiliated. If your users value what you're doing they shouldn't mind you harnessing a little potential that they're leaving on the table. I'd definitely recommend disclosing it though - people will get pissed if you do it silently. Upside is it's totally fire and forget.

Maybe you should also look into reducing the bandwidth cost: check your expiration headers, smaller html/images/css/Javascript, etc. Use Google App engine to get some free traffic, etc.

I'd also suggest looking at Google App Engine. I am moving all my low traffic sites (less than 500 visits a day) to Google App Engine and it's such a great service I wonder why not many people use this.

It has drawbacks but the OP can definitely offload some bandwidth by using App Engine for some parts.

It would help to know what your widget actually did, but here's a freemium pitch: make your widget free for the first X views / mo., and then show ads, unless your user subscribes for $Y / mo.

By show ads, I mean, after some domain has gotten X views, break the widget and show the biggest ad you can.

http://ad.ly/ has been testing an ad network for stream-based applications (those that interact with timeline-based media like Twitter, Facebook, and Myspace's new interface). Ad.ly presented at http://appnationconference.com/ yesterday, both as part of a panel for monetizing the app-based stream, and a break-out panel showing a sample application that includes monetization.

The OneRiot RiotWise Ad API might be right for you:

http://oneriotdevelopernetwork.com/riotwise-api/

You have the option of text ads (typically trending news stories, so they're less spammy than typical ads), small display ads, or full ad units. Additionally, OneRiot's whole approach is real-time/trending information, so it's well within the Twitter-ish sphere.

I have some friends who have used it and reported good CTR and positive (or non-negative) user reactions to the ads.

You could stick "sponsored" tweets into the twitter feed.

That's forbidden by the Twitter API TOS. See http://dev.twitter.com/pages/api_terms Section IV.2.

I had always assumed this just meant that you couldn't inject paid ads into the actual Twitter feed, but reading this in detail you are right, you can't simply mix paid and Twitter content within the display of your app.

Moreover the Twitter API TOS seem to forbid the notion of other sponsorship and advertising for this widget unless Twitter also gets a cut:

> "In cases where Twitter content is the primary basis of the advertising sale, we require you to compensate us [...]"

It sure sounds like this widget is all about Twitter content.

i think this is the easiest, and most likely to be of some success. thought it might alienate those who use the widget, but given the fact you can't sustain it anyway, it's probably not of concern.

edit: looking below, I guess there are restrictions against this. if not the tweet feed, maybe in some other area of the widget.

My company runs a contextual advertising network that could be plugged in with your widget on your publishers pages.

(Can also do non-obtrusive text links which should be more acceptable to your publisher base than banners)

We could really help with the entire monetize using advertising strategy but our service is currently not public.

Contact me offline for details (see profile). (Sorry can't reveal corporate info at this point, apropos of not being public)

Place a text ad for a quality Twitter related affiliate product. Start making commissions. A few I know of...

- http://socialoomph.com

- http://sponsoredtweets.com

- http://tweetadder.com

What space restrictions (banner sizes) are you looking at? I run FeaturedUsers.com and this sounds like something that would be interesting to look at. Not sure how the insertion would work since we do JS insertion based on domains ourselves, but definitely worth looking at.

My contact info is on the site.

Could you add a link on the bottom of the widget to get people to click through to your ad-driven site?

People are only looking at a twitter widget to catch tweets though. If your design is any good, and it seems it must be, then the widget is clearly shown as listing tweets (I can't think what else it would do). Getting people to notice the widget and not then discard it as "just twitter" or use it as twitter and follow the line the tweets send them on seems like too much of an ask IMO.

Adding back links to affiliate pages and selling something using your promoted position in the SERPs seems the most probably model to me, just unfortunately quite scuzzy too.

I am already doing that :)

Time to split test the call to action on the widget to see how you can improve the CTR? :)

Off the top of my head, the value is in the data and analysis that the widget can capture and provide.

Can you track which sites embed the widget? What else can tracked?

It would be helpful to know what the widget is to better come up with useful ideas. But its perfectly understandable that you don't want to share that.

Trying to be vague-ish, but its a customizable widget that displays tweets on a webpage based upon settings the site owner can set in javascript.

Evil way: sell links that you add to the widget to advertisers.

Hard way: upsell "enterprise" versions of the widget to just a few (very expensive) customers.

Good luck!

To add to the idea of enterprise, perhaps you could offer some paid moderation service.

does the widget show a user account's recent tweets? or just the last tweet?

or is it a general/search/trending list?

Any chance you can find a buyer?

perhaps put a flattr link on it?

https://flattr.com/

I'd be very interested in knowing what your site was. My email can be retrieved via my profile/blog. Maybe my similar experience can help you.

In a previous life, I ran tag-board.com, which placed a little javascript widget on your website that let your users interact with each other much like an irc channel. This was before ajax was even coined a term. The site still exists, but it's run down and I don't have anything to do with it anymore.

When I first started it, I never thought more than 100 people would ever use it. At it's peak, there were 600k accounts and tons of traffic that went along with that. At first I didn't care to make money, I was just happy that people used something I created, but when I realized I would need more than a $5/month hosting account, I started looking into options.

The first plan was actually a premium model - hack up a bunch of new features that would only be available to paying users and sell accounts for $20/year. It worked, and might for you as well. At some point, I had over 1,000 paying users.

After awhile, the problem was that hosting costs of the free accounts were eating a good part of the profits on the paid accounts AND I was spending a ton of time on maintainance. I felt responsible for pretty decent uptime/performance/bug-freeness since I had people paying money. A friend of mine pointed out that the amount of time/stress I was putting into this wasn't worth the money. This made a ton of sense to me. At that point, I strongly considered just keeping paying users alive until their year ran out, and then axing the whole thing. I then realized that if I was willing to get rid of all my free users anyway, I might as well try something aggressive - I found an ad network and ran popups on the non-premium accounts. This was ~2002. To my surprise, I got fairly minimal grief, more premium signups, no noticeable drop in usage, and enough income to make it worth my while to keep going. This lasted awhile, but popup revenue slowly dried up - firefox and other browsers started killing the popups. I had handed the whole thing off to someone else well before that happened though.

If I were to do it again, I think I'd do the obvious thing that nobody recommended and just charge everyone for the service, although a much smaller fee (maybe $5/yr). I'd let people sign up for free, use the account for a month, and then switch it over to showing an "ad" of sorts on the owner's website where the widget otherwise would be. Clicking through would let the user interact with the widget, but on the "ad" would also be an option for a user of the website to buy the site owner a subscription to this widget. This would dramatically increase the number of potential customers and would essentially be like a tip jar for the website owner. Users could buy the widget as a gift for a website they like to use. At the same time, I wouldn't try to stop people from just creating a new account and getting another month of free usage, but I'd annoy them forcing them to do this every month.

I don't actually know if this would work, but I think it would. I suspect there would be fewer users. However, as a side benefit, you would know what this is worth to those users, which is a nice feeling. The negative here that scares many people from going this route is that if you suddenly start charging, you really do have to provide a service. You can't go down for 2 weeks. And you have to do a little bit of customer service (emails, refunds, fix bugs) or find someone to do it for you.

$5000 ... you could make some decent money at just a $1 cpm. talk to some ad networks or just sell directly. what kind of sites generate the most pageviews? holy shit, 5 million is a lot of page views. i guess widgets are a great distribution model.

or you could just sell it to websites looking to SEO ^_^ reminds me of the dating site or whatever that had those surveys that people embedded everywhere, that's how they got to #2 or something for "free online dating"

It's not a pageview, it's a widget view. Only the lowest-quality advertisers will pay for that quality and it won't be $1 CPMs. The OP could try putting a (non-compensated) ad on there and measure CTR and how many times the ad even is in the viewable pane (a lot of widgets are below the fold and are never even seen).

The SEO angle won't work-- JavaScript widgets aren't seen by spiders nowadays. The SEO hackers who did what you're referring to added links below the javascript in the code they provided. Downside of this is that if the OP wants to go this route, he can't retroactively add a link to these current users. Unless the link/anchor text relates to widget, Google will ban your ass. The guy who did the online dating thing did in fact get slapped HARD by Google. You can read about it here: http://www.seomoz.org/blog/widgetbait-gone-wild

Agreed, don't go the SEO route - it's not a long-term play.

You got your javascript widget and your noscript tag. You could even stick some regular HTML with a link (possibly an image instead of text) below those. Your users will copy-paste the widget along with your link.

This is a pretty strong SEO technique. You should be doing that already. You don't need to stick unrelated links there, just link back to your Twitter-themed site.

... Which gets him more unmonetized traffic. But yeah, of course he should be doing that.

Note: I wouldn't use the noscript tag. I'd either have a div that was replaced via javascript or a small/light chunk of html below the javascript. Pretty good chance of noscript tags getting dinged by Google given how they are being used.

You're only supposed to have one NOSCRIPT tag per document